Accessing bit fields in floating point values
Posted: Tue Feb 12, 2013 1:52 pm
>from the help line
>How do I access the mantissa and exponent fields of the 32bit word?
>I don't see how to fool the compiler into allowing boolean ops on an FP word.
You fool the compiler (which is how I describe it as well) the same way as in C
C --
i = * (int *) &x; // need type casts in C
BASIC --
i = * (addressof x)
Now you can do bit field tests on the integer. This is how we do detection of
Inf and NaN -- infinity and not-a-number, which happens when you exceed the
E+/-38 range of floating points.
>How do I access the mantissa and exponent fields of the 32bit word?
>I don't see how to fool the compiler into allowing boolean ops on an FP word.
You fool the compiler (which is how I describe it as well) the same way as in C
C --
i = * (int *) &x; // need type casts in C
BASIC --
i = * (addressof x)
Now you can do bit field tests on the integer. This is how we do detection of
Inf and NaN -- infinity and not-a-number, which happens when you exceed the
E+/-38 range of floating points.